Search it
Production enterprise search over your own content
Faceted, multilingual, typo-tolerant search on the engine you already run (Apache Solr, Elasticsearch or an embedded Lucene index) behind one query API. No data leaves your infrastructure.
One query API, three engines
Point each search site at Solr, Elasticsearch or the built-in Lucene index and query them all through the same REST API. Swap engines per site without touching your front-end; the search-engine plugin layer abstracts the difference.
Faceted, multilingual & typo-tolerant
Facets, locale-aware analyzers (English, Portuguese, Spanish, Catalan), autocomplete, spell-check and currency/number localization ship in the box.
Schema as code
Describe a site's fields in a versioned manifest and POST it. Turing converges the live schema to match, with a diff planner that flags breaking changes before they run. No schema yet? Derive a draft from sample documents.
Hybrid ranking that fuses keyword + vector
Opt into hybrid ranking and Turing reorders results with Reciprocal Rank Fusion over BM25 and per-site vector scores, so semantic matches rank well and exact matches still count.
Coverage you can see
A field-coverage view shows what percentage of your documents actually populate each field, so you catch a half-indexed catalog before your users do.
curl -X POST https://your-host/api/sn/products/search \
-H "Content-Type: application/json" \
-d '{
"q": "wireless headphones",
"rows": 20,
"fq": ["brand:Acme", "price:[100 TO 300]"],
"_setlocale": "en_US"
}'Questions teams ask
- Do I need Solr or Elasticsearch to start?
- No. A site can use the embedded Lucene index that runs inside Turing, so a single container is a working search stack. Move the site to Solr or Elasticsearch later; the query API and your front end stay the same.
- How does hybrid keyword and vector ranking work?
- It is opt-in per site. With hybrid ranking on, Turing runs the keyword query and a vector query over the site's own embeddings, then merges the two lists with Reciprocal Rank Fusion. Sites that leave it off keep plain keyword ranking.
- Can I keep my synonyms when I move from Algolia?
- Yes. Synonyms are native, query-time and need no reindex, and the admin can import an Algolia synonym export. Multi-word synonyms such as "home office" expand as phrases.
- How do I define a site's fields?
- Write them in a versioned manifest and POST it. Turing plans the change, shows what would break, and converges the live schema. With no schema yet, it can draft a manifest from sample documents for you to review.